Working local research edition · 316,926 current transaction records · 1,888 filers · snapshot orestar-20250101-20260927-v1. Descriptive findings are reproducible; expanded source review and several research tracks remain incomplete. Causal interpretations are not claimed.
Three stories emerging from the records.
One reported gift changes the September story
September 1–27 reports $20,056,194.92 in cash contributions. Transaction 5816290—the $10 million Carpenters contribution to Building a Stronger Oregon—is 49.9% of that total. Excluding that one record leaves $10,056,194.92, versus $13,697,269.57 in the full month of August.
An aggregate fundraising surge can reflect a single institutional funding event rather than a broad-based increase in participation.
September is incomplete, reporting can arrive later, and the comparison is not seasonally adjusted. No claim about acceleration, spending, donor intent or electoral effect. Other campaigns can accelerate even when an aggregate comparison is dominated by one receipt. Gross receipts still include committee transfers.
Evidence: monthly.csv · largest-cash-records.csv
Similar 2026 fundraising masks sharply different reported cash positions
Kotek’s committee reports $10,293,083.12 in 2026 cash contributions and Drazan’s $9,764,922.48. Their retrieved official summaries report closing cash of $8,376,671.93 and $964,650.50, respectively—a 8.68-to-one ratio.
Kotek started 2026 with $2,432,342.22 and reports $4,353,542.97 in cash expenditures. Drazan started with $982,810.75 and reports $9,774,591.62. Both starting resources and spending distinguish their positions.
These are reported—not bank-verified—balances. Drazan’s official cash payments exceed the fixed snapshot by $800.00; that unresolved difference is retained. Cash balances do not measure unpaid commitments, campaign quality or election probability. Earlier spending may already have purchased advertising, organization or other useful capacity. A lower balance is not itself a weaker campaign or proof of poor management.
Evidence: account-summaries.csv · enrichment.json
Public matching is the majority of reported cash for all 13 linked certified candidates
Classified matching-fund receipts supply 66.1% of the combined $2,062,510.81 in cash contributions reported by the 13 linked Portland certified candidates. Every one of the 13 is above half; individual shares range from 50.9% to 82.4%.
In this group, a gross fundraising ranking is substantially a public-financing ranking. It cannot be read as a ranking of private fundraising alone.
Not all Portland candidates, not a causal comparison of participants with nonparticipants, and not an exact City disbursement-level match. Unclassified receipts are not certified private money. The City and fixed ORESTAR observations differ in timing and coverage. Public matching may reflect qualifying participation, but these totals alone do not establish broader or more representative donor support.
Evidence: portland-public-financing.csv · public-matching-receipts.csv · committee-race-crosswalk.json
1. Start with what the dollars mean.
The snapshot reports $121,794,170.02 in cash contributions and $91,270,955.55 in cash expenditures. Neither number is a measure of “money spent to win elections.” Contributions include internal committee transfers and public funding. Expenditures include transfers, reimbursements, overhead and activity outside candidate campaigns. Reproduce the accounting totals.
Another $10,947,490.20 appears as in-kind support. Loans received total $1,267,729.83. Unpaid accounts and personal expenses submitted for reimbursement are obligations, not a second set of cash purchases. Their later payments must be connected through the underlying records before reconstructing debt.
Official account summaries are now available for 25 selected committees, covering 41 committee-years. 39 agree with the snapshot’s cash receipts, cash payments and net cash change; 2 have retained differences. This is targeted coverage, not a statewide balance audit. Balances and reconciliation differences.
This distinction changes the substantive question. A campaign can have substantial reported support without receiving the same amount in spendable cash. A committee can move money to another committee without creating a new dollar in Oregon politics.
2. The largest recipients account for a large share—but not a single kind of political power.
The three largest filing committees together report $34,383,392.21 in cash contributions: 28.2% of this snapshot’s reported cash-contribution dollars. Two are candidate committees, Friends of Tina Kotek and Friends of Christine Drazan. The third, Building a Stronger Oregon, is a political action committee. Ranking them together describes receipts, not comparable election contests. Committee ledger.
A single September 16, 2026 contribution reported by Building a Stronger Oregon names North Coast States Regional Council of Carpenters and records $10 million. That one record is 8.2% of all reported cash contributions in this period. It is a reported transaction, not a finding that those dollars were spent or that a recipient was obligated to act for the contributor. Inspect transaction 5816290 and that day’s evidence.
Among identified cash-source record groups—including committees—the ten largest account for 21.5% of identified dollars. The Gini coefficient is 0.905. This is a concentration measure over provisional groups, not a count of verified people. A name change can split a source; an unresolved shared address can still conceal differences. Source groups and their evidence.
Removing the largest visible source lowers the Gini to 0.896 and the top-ten share to 14.7%. Aggressive same-name/type grouping yields a Gini of 0.913. The pattern remains uneven, but no scenario is treated as a verified identity map. All eight sensitivity scenarios.
3. A small transaction is not necessarily a small donor.
The source contains 32,258 contribution-family records totaling $12,084,155.64 under miscellaneous or anonymous labels. Those records preserve money while withholding an individually identifiable source. Treating “Miscellaneous Cash Contributions $100 and under” as one donor would invent a giant donor connected to unrelated campaigns. We do not do that. Inspect every disclosure category.
Transactions of $100 or less are 69.5% of nonaggregate cash rows but 3.1% of their dollars. Transactions over $1,000 are 4.9% of these rows and 81.7% of their dollars. This comparison counts transactions, not people. Exact size-band counts and amounts.
For identified groups, gifts are accumulated within each committee before calculating small-source measures. A sequence of $25 gifts is different from one $25 giver. But the reported aggregate categories cannot be distributed back to people, and an identified person’s visible total may omit earlier unitemized giving. “Visible cumulative giving of $100 or less” is therefore an observation rule, not a certification that someone is a small donor.
Campaign dependency is measured in two ways: concentration among visible groups, and a sensitivity range that treats unidentified money as extremely dispersed or concentrated with the largest visible source. Neither endpoint is an estimate of the truth. Together they show how much the conclusion depends on disclosure. Committee dependency and disclosure sensitivity.
4. Follow transfers without pretending to follow individual dollars.
A strict reconciliation finds 1,454 pairs totaling $7,796,853.20: the sender’s payment and recipient’s receipt have reciprocal committee IDs, identical amounts and the same transaction date, with a unique candidate on each side. Both records remain in the evidence. Matched transfer ledger.
This is a conservative view. Another 2,419 candidate pairs meet the broader seven-day rule but fail the strict same-day uniqueness test. Some are ordinary timing differences; others are ambiguous recurring amounts. For example, the $1,208,352.82 nurses’ PAC payment on May 15, 2025 and receipt on May 16 are deliberately outside the strict total. This counterexample shows why “unmatched” does not mean suspicious. Review the alternative matches.
A maximum one-to-one matching within those seven-day candidates can reconcile $17,927,165.51 across 3,847 pairs. This is a possible reconciliation inside the candidate set, not a verified transfer total or an upper bound on every internal transfer. Matching assumptions and sensitivity results.
After subtracting strict matched circulation, $113,997,316.82 remains. That is not a verified total of money entering from outside: unmatched internal transfers and uncertain origins still remain. Because funds are commingled, a path from a contributor through a committee to a vendor establishes a possible route—not proof that the contributor’s dollar paid that vendor.
5. The spending record is also a services economy.
The largest noncommittee payee record groups are dominated by media, organizing, management and campaign operations. But a payment to an intermediary is not necessarily that firm’s revenue: media buying, payroll and reimbursement can pass money onward. The explorer therefore labels these amounts “cash payments,” not profits or vendor earnings. Payee record groups.
The purpose codes also describe bundles. A payment tagged both management and advertising cannot be counted in full under each category. We retain the exact code combination until an actual allocation is available. Unspecified purposes stay visible. Spending-purpose bundles.
6. Timing and geography reveal activity—not motives.
Each contribution retains transaction date and the current version’s filing date. Amended filings can appear long after the underlying event; that lag is not necessarily how long voters waited for the original disclosure. The source snapshot excludes superseded versions, so a historical “what voters knew” reconstruction needs more records. Filing-lag distributions.
State and city summaries use reported contributor locations. They do not establish legal residence, voting eligibility or membership in a council district. Missing locations remain in the denominator. No residential-address map is published. State-level contribution geography.
7. Portland asks a different financing question.
Portland’s matching program means campaign resources are not synonymous with private donor giving. Public matching payments must be separated from private contributions, and the City’s distribution records must be reconciled to ORESTAR before either is added to a total. A City disbursement and a committee’s report of receiving it are two descriptions of one payment.
The archived City program page reports $1,542,435.00 distributed among 13 certified candidates. The fixed ORESTAR snapshot contains $1,363,713.00 in reviewed matching-fund receipt records for those committees, a difference of $178,722.00. These are overlapping observations, not two pots of money. Reporting timing, amendments and incomplete source coverage remain possible explanations; the difference is not an allegation. Candidate-by-candidate reconciliation.
The City publishes program certification and matching-fund distributions for the 2026 election. Participation can be compared descriptively, but participating candidates select into the program and compete under different conditions. An observed difference in donor breadth is not, by itself, a causal effect of public financing. Official 2026 program status.
Open the reviewed Portland race comparisons. Profiles without a verified committee crosswalk are marked missing, not assigned zero fundraising. The November 3, 2026 election is pending and receives no outcome analysis.
8. What does the money change?
This record establishes reported resources and financial connections. It cannot alone establish why an official voted a particular way, whether campaigns coordinated, or whether a contributor bought influence. Those questions require dated government records, verified identities, relevant comparisons and competing explanations.
The network analysis asks a narrower question: do selected campaigns share more visible donor groups than would be expected from their donor counts and the groups’ propensity to give across campaigns? Its degree-preserving permutation tests keep those features fixed. Even an unusually strong overlap is not evidence of illegal coordination. All 435 exploratory pair tests; assumptions, seeds and diagnostics.
The defensible story is already consequential: money is highly uneven, a meaningful amount circulates through committees, and the record’s ability to identify people is not the same as its ability to count dollars. The remaining questions are preserved, including null results, counterexamples and things this edition cannot yet establish.